Contrast in levels of morphological versus molecular divergence between closely related Eurasian species ofPlatanthera(Orchidaceae) suggests recent evolution with a strong allometric component

We consider the conceptual relationship between pollinator specfificity, genetic isolation and species delimitation, critiquing Darwin’s (1877) hypothesis that differential placement of pollinia on the probosces and eyes of sphingid moths explained the divergence of the Eurasian orchids Platanthera bifolia and P. chlorantha — species that have similar geographic distributions, habitat preferences and flowering times. Subsequent workers have developed Darwin’s initial hypothesis into an oft-cited model system for co-evolutionary speciation via a prezygotic barrier. The four datasets gathered from the two species were: (a) morphometric data from 139 individuals of 21 populations in southern England, supported by SEM study of representative flowers; (b) spur-length measurements gathered by many fieldworkers from 3070 plants sampled across western Europe; (c) DNA sequences from two representative individuals of each species from nuclear ribosomal ITS and eight plastid regions; and (d) nrITS sequence data for 50 individuals of the two species and their close relatives in Eurasia. Platanthera bifolia and P. chlorantha show a strong allometric relationship approximating a 2∶3 ratio in mean values for most measurements. The few characters showing greater deviations collectively reflect some but not all aspects of the classical adaptive scenario; moreover, they may be genetically linked. Those characters are sufficient to reliably distinguish between the two species and identify occasional hybrids. Spur length of both species shows a latitudinal gradient, strongest within the British Isles, that decreases northward by an average of ca. 2% per 100 km. In contrast with this substantial morphological divergence, only one base-pair difference was detected in c.9 kb of rapidly evolving nuclear and plastid DNA, though inclusion of closely related taxa from the Mediterranean region, Macaronesian islands and eastern Asia increased to seven the number of subtly distinct ITS alleles recovered from the aggregate. Most of the phenotypic variability observed, both within and between P. bifolia and P. chlorantha, is encompassed within a single allometric cline that conflates — in order of decreasing influence — ontogenetic factors, environmental (epigenetic) influences, and taxonomic distinction. Phenotypic divergence between the two species is less than is generally supposed, and is hypothesised to reflect a simple genetic control that radically increases stigma size in P. chlorantha. Our failure to detect species-specific genetic differences between the two allogamous putative species indicates (but cannot conclusively demonstrate) that extensive gene-flow occurs between them, suggesting that previous assertions of pollinator specificity have been greatly exaggerated. This homogeneity could represent hybridisation through secondary contact of formerly distinct species, but this theory is contradicted by the low allelic diversity of the putative species, together with their similar geographic distributions and habitat preferences. Rather, we suspect that these and other closely related taxa represent a relatively early stage of speciation, when phenotypic divergence (which typically reflects minute changes in the regulation of phenotypically expressed genes) inevitably precedes more widespread divergence in the various genic regions that are routinely used in molecular phylogenetics. This phenotypically overt but genotypically cryptic phase of speciation, here termed the genetic divergence lag (GDL), renders such incipient species immune to DNA barcoding. Many incipient species may never achieve a level of genetic isolation sufficient to escape the GDL. In the present case, the incipient speciation may have occurred sympatrically and may (thus far) have led only to stabilised polymorphism.